Intelligent air cutting plate processing method and system

Abstract

The application provides a processing method and system for intelligent air cutting of plates, and belongs to the technical field of laser plate processing. The method comprises the following steps: obtaining plate basic information, calling an air cutting process database to match initial air cutting parameters; purifying compressed air and using it as auxiliary gas; loading a cutting track and collecting multi-channel sensing signals in the cutting process, including plasma light signals, signals of the distance between a cutting head nozzle and the plate, and gas pressure feedback signals; inputting the multi-channel sensing signals into a process adaptive model to output laser power compensation, feed speed compensation, and gas pressure compensation for dynamic correction; driving the cutting head to complete plate cutting based on the corrected parameters; collecting cutting images for quality determination, and recording process data and defect data to the database and triggering model iteration update when defects exist. The application realizes intelligent adaptive regulation and control of the whole process of air laser cutting, improves cutting quality consistency, and reduces manual dependence and gas cost.

Description

Technical Field

[0001] This application belongs to the field of laser sheet processing technology, specifically relating to a method and system for intelligent air-cutting sheet processing. Background Technology

[0002] When cutting sheet metal parts with fiber lasers, the mainstream solution uses nitrogen as an auxiliary gas to obtain a bright, oxygen-free cut. However, the continuous purchase of nitrogen leads to high costs. Air cutting solutions that use compressed air instead of nitrogen can significantly reduce gas costs. However, current air cutting technology is still immature, mainly due to the following issues: process parameters are highly dependent on manual experience for adjustment, poor adaptability to fluctuations in sheet thickness and gas source pressure, and a tendency to overheat or slag when cutting corners, small holes, and other features; at the same time, there is a lack of effective control over the water and oil content of compressed air, resulting in significant issues of lens contamination and deterioration in cutting quality.

[0003] To address the aforementioned issues, some solutions have attempted to incorporate process databases for parameter matching. For example, they utilize cutting status acquisition devices to obtain processing status information, and then expert systems call upon knowledge bases for parameter optimization. However, these solutions are rule-based expert systems, where parameter optimization relies on matching existing rules in the knowledge base. When no matching rules exist, manual intervention is required, making it impossible to achieve dynamic closed-loop control based on real-time sensor signals. Furthermore, the acquired signals are mostly quality feedback information from the cutting surface, without real-time fusion sensing of multi-source signals such as plasma optical signals, capacitor adjustment signals, and gas pressure signals. Other solutions employ AI-based cutting status monitoring and parameter optimization strategies, such as using convolutional neural networks for cutting status assessment and multi-objective optimization algorithms for process parameter decisions. However, these solutions use a general laser cutting intelligent control framework and are not specifically designed for the unique process requirements of using compressed air instead of nitrogen. They do not address compressed air purification or propose differentiated energy compensation strategies for geometric features such as corners, small holes, and narrow slits.

[0004] Therefore, in existing air laser cutting, there is a lack of real-time closed-loop control methods for the cutting status, multi-source signals are not effectively integrated and utilized, the quality of compressed air source is difficult to control online, and process parameters still rely on repeated manual adjustments, making it impossible to achieve automatic iterative optimization based on cutting quality feedback. Summary of the Invention

[0005] In a first aspect, embodiments of this application provide a method for processing intelligent air-cutting sheet metal, comprising the following steps: S1. Obtain the basic information of the board to be processed, wherein the basic information of the board includes at least the board material and the board thickness; S2. Call the air cutting process database and match the initial air cutting parameters according to the basic information of the plate. The initial air cutting parameters include at least the laser power, cutting feed speed, focal height and compressed air pressure. S3. Start the air compressor supply device to purify the compressed air and use the purified compressed air as the laser cutting auxiliary gas; S4. Load the cutting trajectory and execute the cutting, and collect multiple sensor signals in real time during the cutting process. The multiple sensor signals include plasma light signals, the distance signal between the cutting head nozzle and the plate, and gas pressure feedback signals. S5. Input the multi-sensor signals into the trained process adaptive model, the process adaptive model outputs at least one of laser power compensation, feed speed compensation and air pressure compensation, and uses the compensation to dynamically correct the current air cutting parameters. S6. Based on the corrected air cutting parameters, the laser cutting head is driven to move along the loaded cutting trajectory, and the plate is cut using purified compressed air; S7. Collect the cut image after the cutting is completed and judge the quality. If a defect is judged, record the process data and defect data of the current cutting process to the air cutting process database to trigger the iterative update of the process adaptive model.

[0006] Secondly, embodiments of this application also provide a processing system for intelligent air-cutting sheet metal, comprising: The board material information acquisition module is used to acquire basic information about the board material to be processed, which includes at least the board material and the board thickness. The process parameter matching module is used to call the air cutting process database and match the initial air cutting parameters according to the basic information of the plate. The initial air cutting parameters include at least laser power, cutting feed speed, focal height and compressed air pressure. The compressed air supply and purification module is used to start the air compressor supply device, purify the compressed air, and use the purified compressed air as the auxiliary gas for laser cutting. The multi-signal sensing module is used to load the cutting trajectory and execute the cutting, and to collect multiple sensing signals in real time during the cutting process. The multiple sensing signals include plasma optical signals, the distance signal between the cutting head nozzle and the plate, and gas pressure feedback signals. The intelligent parameter control module is used to input multi-sensor signals into the trained process adaptive model. The process adaptive model outputs at least one of laser power compensation, feed speed compensation and air pressure compensation, and uses the compensation to dynamically correct the current air cutting parameters. The laser cutting execution module is used to drive the laser cutting head to move along the loaded cutting trajectory based on the corrected air cutting parameters, and to cut the sheet material using purified compressed air; The vision inspection module is used to acquire images of the cut surface after the cutting is completed and to determine the quality. If a defect is determined, the process data and defect data of the current cutting process are recorded in the air cutting process database to trigger the iterative update of the process adaptive model.

[0007] As can be seen from the above technical solutions, this application has the following advantages: The intelligent air cutting method and system provided in this application reduce the gas cost of sheet metal cutting by replacing nitrogen with compressed air. Simultaneously, it constructs a full-process intelligent adaptive control system, using multi-source sensor signal fusion to perceive the cutting status, combined with strict compressed air purification control, solving the problems of lens contamination and cut quality deterioration. For the complex geometric features of corners and small holes, it achieves differentiated energy compensation adjustment, improving overheating and slag formation. Furthermore, through closed-loop feedback of cut quality and self-iterative updates of the process adaptive model, it can continuously optimize process parameters, overcoming the dependence on human experience and sensitivity to operating condition fluctuations in traditional air cutting, achieving low-cost, high-stability automated intelligent cutting. Attached Figure Description

[0008] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a schematic flowchart of the processing method for the intelligent air-cutting sheet material of the present invention.

[0010] Figure 2 This is a schematic diagram of the intelligent air-cutting plate processing system of the present invention. Detailed Implementation

[0011] The various embodiments of this disclosure will be described more fully in the detailed steps of the processing method for intelligent air-cutting sheet metal described below. This disclosure may have various embodiments, and adjustments and changes may be made therein. However, it should be understood that there is no intention to limit the various embodiments of this disclosure to the specific embodiments disclosed herein, but rather this disclosure should be understood to cover all adjustments, equivalents, and / or alternatives falling within the spirit and scope of the various embodiments of this disclosure.

[0012] This embodiment provides a method for processing sheet metal using intelligent air cutting. Through intelligent control of the entire process, it solves the problems of unstable air cutting quality, reliance on manual labor, and high costs.

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] Please see Figure 1 The diagram shows a flowchart of a processing method for intelligent air-cutting sheet metal in a specific embodiment. The method includes the following steps: S1. Obtain the basic information of the board to be processed, wherein the basic information of the board includes at least the board material and the board thickness; It should be noted that by obtaining the material, thickness, and surface condition of the sheet metal, a data foundation is provided for the accurate matching of process parameters; S2. Call the air cutting process database and match the initial air cutting parameters according to the basic information of the plate. The initial air cutting parameters include at least the laser power, cutting feed speed, focal height and compressed air pressure. It should be noted that the initial air cutting parameters are automatically matched by calling the process database based on the basic information of the sheet material, which reduces manual experience-based debugging and improves the efficiency and accuracy of parameter setting. S3. Start the air compressor supply device to purify the compressed air and use the purified compressed air as the laser cutting auxiliary gas; It should be noted that multi-stage purification effectively removes moisture and oil from compressed air, ensuring air quality and extending the service life of optical components such as lenses and nozzles. S4. Load the cutting trajectory and execute the cutting, and collect multiple sensor signals in real time during the cutting process. The multiple sensor signals include plasma light signals, the distance signal between the cutting head nozzle and the plate, and gas pressure feedback signals. It should be noted that by acquiring plasma light signals, the distance between the cutting head nozzle and the plate, and gas pressure feedback signals in real time, comprehensive perception of multi-source state information during the cutting process is achieved, providing a real-time data basis for dynamic control. S5. Input the multi-sensor signals into the trained process adaptive model, the process adaptive model outputs at least one of laser power compensation, feed speed compensation and air pressure compensation, and uses the compensation to dynamically correct the current air cutting parameters. It should be noted that by inputting multi-source sensor signals into the process adaptive model and outputting compensation amounts in real time for dynamic correction, closed-loop adaptive control of cutting parameters is achieved, thereby improving the consistency of cutting quality. S6. Based on the corrected air cutting parameters, the laser cutting head is driven to move along the loaded cutting trajectory, and the plate is cut using purified compressed air; It should be noted that the cutting head movement is driven by the corrected parameters, which ensures a stable and high-quality cutting process under complex trajectories and fluctuating working conditions. S7. Collect the cut image after the cutting is completed and judge the quality. If a defect is judged, record the process data and defect data of the current cutting process to the air cutting process database to trigger the iterative update of the process adaptive model. It should be noted that by collecting cut images for quality assessment and recording defect data, a quality closed-loop feedback mechanism was constructed, which triggered the model's self-iteration to continuously improve the cutting process level.

[0015] This embodiment achieves reduced cutting costs and consistent and stable cutting quality by using intelligent adaptive control throughout the entire process and replacing nitrogen with compressed air.

[0016] Furthermore, as a refinement and extension of the specific implementation of the above embodiments, in order to fully illustrate the specific implementation process of this embodiment, another intelligent air-cutting method for processing sheet metal is provided, which includes the following steps: S1. Obtain the basic information of the board to be processed, wherein the basic information of the board includes at least the board material and the board thickness; The specific steps of step S1 are as follows: S11. Input the material and thickness of the sheet to be processed through the human-computer interaction interface or import it through the upstream system; the sheet material includes, but is not limited to, carbon steel and stainless steel, and the sheet thickness ranges from 1mm to 10mm. S12. Obtain surface condition information of the board material through a visual sensor or laser scanner, wherein the surface condition information of the board material includes at least one of surface roughness, oxide layer thickness and surface flatness; In this embodiment, the material, thickness, and surface condition of the board are combined into a basic information vector of the board, which serves as the input condition for subsequent process database retrieval and matching. By acquiring the above multi-dimensional basic information, the matching of initial process parameters can be more accurate, avoiding blind trial cutting due to batch differences in the board. S13. Combine the board material, board thickness, and board surface condition into a basic information vector of the board; S2. Call the air cutting process database and match the initial air cutting parameters according to the basic information of the plate. The initial air cutting parameters include at least the laser power, cutting feed speed, focal height and compressed air pressure. The specific steps of step S2 are as follows: S21. Using the basic information of the sheet metal as the search criteria, query the air cutting process database; the air cutting process database stores several historical process records, each historical process record includes the sheet metal material, sheet metal thickness, sheet metal surface condition and corresponding optimal air cutting parameters, the optimal air cutting parameters include at least laser power, cutting feed speed, focal height, compressed air pressure and nozzle type; S22. Retrieve historical process records in the air cutting process database that match the current sheet material and sheet thickness; If multiple historical process records are matched, the air cutting parameters corresponding to the historical process record that is closest to the current surface condition of the board are selected as the initial air cutting parameters. If no matching record is found, the default parameter group will be used as the initial air cutting parameters. S23. Output the initial air cutting parameters obtained from the matching; S3. Start the air compressor supply device to purify the compressed air and use the purified compressed air as the laser cutting auxiliary gas; The specific steps of step S3 are as follows: S31. Start the air compressor to generate compressed air and send the compressed air into the air tank for buffering and pressure stabilization; S32. After buffering and stabilizing, the compressed air is sequentially passed into an oil-water separator for preliminary oil-water separation, into a refrigerated dryer for drying, and into a multi-stage filter with progressively increasing filtration precision for filtration and purification; the multi-stage filter includes at least a primary filter and a secondary filter, wherein the filtration precision of the secondary filter is higher than that of the primary filter. The primary filter has a filtration accuracy of 3 μm, and the secondary filter has a filtration accuracy of 0.01 μm. S33. The pressure of the purified compressed air is stabilized by a pressure regulating valve to control the dew point and oil concentration of the purified compressed air within preset values ​​(e.g., dew point ≤ 3℃, oil concentration ≤ 0.01mg / m³). S34. The purified compressed air after pressure stabilization is delivered to the laser cutting head through the air supply pipeline as auxiliary gas, and the real-time pressure value of the air supply pipeline is output as a gas pressure feedback signal. The purification process ensures the quality of the air source entering the cutting head, avoiding contamination and damage to optical components by water and gasoline. S4. Load the cutting trajectory and execute the cutting, and collect multiple sensor signals in real time during the cutting process. The multiple sensor signals include plasma light signals, the distance signal between the cutting head nozzle and the plate, and gas pressure feedback signals. The specific steps of step S4 are as follows: S41. Read the G-code file of the cutting trajectory of the part to be processed and parse it to obtain the preset motion trajectory of the laser cutting head; S42. Start laser cutting processing and simultaneously activate multiple sensors for sensing; S43. The intensity of the plasma light signal generated during the cutting process is acquired in real time by a plasma sensor, wherein the intensity of the plasma light signal characterizes the melting and vaporization state of the material in the cutting area; S44. The capacitance value between the cutting head and the plate is detected in real time by a capacitance height sensor, and the real-time distance signal of the cutting head is calculated based on the capacitance value. The real-time distance signal represents the following height between the nozzle of the cutting head and the surface of the plate. The focal height is determined by the configuration of the optical lens group inside the cutting head and is set as an independent process parameter in step S2. S45. Real-time acquisition of gas pressure feedback signals in the gas supply pipeline via a gas pressure transmitter; S46. The real-time collected plasma optical signal, the distance signal between the cutting head and the plate and the gas pressure feedback signal are combined into a multi-channel sensing signal vector and transmitted to the intelligent parameter control module at preset sampling intervals as input to the process adaptive model described in step S5. In this embodiment, the real-time acquisition of multiple sensor signals is achieved in the following way: First, the cutting trajectory G-code file of the part to be processed is read and parsed to obtain the preset motion trajectory of the laser cutting head, including the geometric feature information of the trajectory such as corner position, hole position and slit position, etc. Simultaneously with the start of laser cutting, multiple sensors are activated to collect data. The intensity of the plasma light signal generated during the cutting process is acquired in real time by a plasma sensor. The intensity of the plasma light signal characterizes the melting and vaporization state of the material in the cutting area. The plasma sensor can be a photodiode or a spectrometer, with a acquisition band range of 300-800 nm and a sampling frequency of not less than 1 kHz. The capacitance value between the cutting head and the plate is detected in real time by a capacitance adjustment sensor, and the real-time distance signal between the cutting head nozzle and the plate surface is calculated based on the calibration relationship between the capacitance value and the distance. The real-time distance signal represents the following height between the cutting head nozzle and the plate surface. The focal height is determined by the configuration of the internal optical lens group of the cutting head and is set as an independent process parameter in step S2, and is not measured by the capacitance adjustment sensor. The detection accuracy of the capacitance adjustment sensor is not less than ±0.01mm, and the sampling frequency is not less than 500Hz. The gas pressure feedback signal in the gas supply pipeline is collected in real time by a gas pressure transmitter. The detection accuracy of the gas pressure transmitter is not less than ±0.01MPa and the sampling frequency is not less than 200Hz. The real-time acquired plasma light signal, the distance signal between the cutting head nozzle and the plate, and the gas pressure feedback signal are combined into a multi-channel sensing signal vector and transmitted to the intelligent parameter control module at preset sampling intervals (e.g., 10-50ms) as input for the subsequent process adaptive model. S5. Input the multi-sensor signals into the trained process adaptive model, the process adaptive model outputs at least one of laser power compensation, feed speed compensation and air pressure compensation, and uses the compensation to dynamically correct the current air cutting parameters. The specific steps of step S5 are as follows: S51. Construct a process adaptive model, wherein the process adaptive model is a neural network model trained with historical cutting process data, wherein its input layer receives the multi-channel sensor signals, and its output layer outputs laser power compensation, feed speed compensation and air pressure compensation. The model construction and inference process is as follows: Historical cutting process data covering different sheet materials (carbon steel, stainless steel), different sheet thicknesses (1mm~10mm), and different cutting conditions are pre-collected, including multi-channel sensor signal sequences at each time point and corresponding quality evaluation labels, forming a training dataset; a supervised learning approach is adopted, using multi-channel sensor signal sequences as input features and the compensation amount that optimizes cutting quality as the label to train the neural network model; the model uses a multilayer perceptron (MLP) or long short-term memory (LSTM) network structure, where the LSTM structure is suitable for processing time-series signals; after training, the model is deployed to the intelligent parameter control module; S52. Input the multi-channel sensor signals that are collected and transmitted in real time into the process adaptive model. The process adaptive model performs forward inference calculation and outputs at least one of the laser power compensation, feed speed compensation and air pressure compensation at the current moment. S53. The compensation amount output by the process adaptive model is superimposed on the current air cutting parameters to dynamically correct the current air cutting parameters:

[0017]

[0018]

[0019] in, This is the corrected real-time laser power. The current laser power, This is the laser power compensation amount; This is the corrected real-time cutting feed rate. This represents the current cutting feed rate. This is the feed rate compensation amount; This is the corrected real-time compressed air pressure. The current compressed air pressure, This is the air pressure compensation amount; S54. Output the corrected air cutting parameters; In step S5, geometric features on the cutting trajectory are also identified, including at least one of corners, small holes, and narrow slits; When the geometric features are identified, the corrected air cutting parameters are adjusted for energy compensation. The energy compensation regulation specifically includes: Based on the type and size of the geometric features, the corresponding power compensation coefficient and speed reduction coefficient are determined respectively; The current laser power is corrected using a power compensation factor:

[0020] in, This represents the laser power after energy compensation. The current laser power, The power compensation coefficient corresponding to the geometric feature type; The current cutting feed rate is corrected using a speed reduction coefficient:

[0021] in, This refers to the cutting feed rate after energy compensation. This represents the current cutting feed rate. The velocity deceleration coefficient corresponding to the geometric feature type; When the geometric feature is a corner, the power compensation coefficient and speed deceleration coefficient are determined based on the interior angle of the corner. The smaller the interior angle of the corner, the smaller the power compensation coefficient and the smaller the speed deceleration coefficient. When the interior angle of the corner is greater than or equal to 170°, it is determined to be a straight line, and both the power compensation coefficient and the speed deceleration coefficient are 1; compensation is turned off. In this embodiment, the value range of the power compensation coefficient is 0.70 to 1.25. When the calculated value exceeds this range, it is clamped at the boundary value. For example, the speed deceleration coefficient is 0.40 when the interior angle of the corner is 15° to 30°, 0.55 when the interior angle of the corner is 30° to 60°, 0.70 when the interior angle of the corner is 60° to 90°, 0.85 when the interior angle of the corner is 90° to 135°, and 1.00 when the interior angle of the corner is 135° to 180°. When the geometric feature is a small hole, the power compensation coefficient and speed deceleration coefficient are determined according to the diameter of the small hole. The smaller the diameter of the small hole, the smaller the power compensation coefficient and the smaller the speed deceleration coefficient. When the diameter of the small hole is greater than a preset diameter threshold, both the power compensation coefficient and the speed deceleration coefficient are 1, and compensation is turned off. In this embodiment, the value range of the small hole power compensation coefficient is 0.65 to 1.30. For example, the value of the small hole speed deceleration coefficient is as follows: when the hole diameter is 0.5 to 1.0 mm, the speed deceleration coefficient is 0.35; when the hole diameter is 1.0 to 2.0 mm, the speed deceleration coefficient is 0.50; when the hole diameter is 2.0 to 3.0 mm, the speed deceleration coefficient is 0.75; and when the hole diameter is greater than 3.0 mm, the speed deceleration coefficient is 1.00. When the geometric feature is a narrow slit, the power compensation coefficient and speed deceleration coefficient are determined according to the slit width. The smaller the slit width, the smaller the power compensation coefficient and the smaller the speed deceleration coefficient. When the slit width is greater than a preset width threshold, both the power compensation coefficient and the speed deceleration coefficient are 1, and compensation is turned off. In this embodiment, the value range of the slit power compensation coefficient is 0.60 to 1.20. For example, the value of the slit speed deceleration coefficient is as follows: when the slit width is 0.4 to 0.8 mm, the speed deceleration coefficient is 0.60; when the slit width is 0.8 to 1.5 mm, the speed deceleration coefficient is 0.80; when the slit width is greater than 2.5 mm, the speed deceleration coefficient is 1.00. The power compensation coefficient has a preset value range. When the calculated value exceeds the value range, the power compensation coefficient is clamped to the boundary value of the value range. The power compensation coefficient of the corner is 0.70 to 1.25, the power compensation coefficient of the small hole is 0.65 to 1.30, and the power compensation coefficient of the narrow slit is 0.60 to 1.20. In some embodiments, the power compensation coefficient and velocity deceleration coefficient are both smoothed during entry into and exit from the geometric feature, so that the laser power and cutting feed speed change gradually rather than abruptly within the regions before and after the geometric feature; the smoothing process employs a first-order low-pass filter.

[0022] in, The compensation coefficient at the current moment. This is the compensation coefficient from the previous moment. The target compensation coefficient is... The smoothing coefficient and The smaller α is, the smoother the transition; the smooth transition time Ts can be set to 20–50 ms; simultaneously, compensation coefficients are gradually loaded within a preset trajectory interval before reaching the geometric feature (e.g., 5–10 mm from the feature point) to achieve early triggering; after leaving the feature region, the compensation coefficients gradually return to 1 after passing through the smooth transition interval to achieve delayed exit; in addition, the maximum power change amplitude within a single sampling period is set. (For example, not exceeding 10% of the current power), and each power change should not exceed [a certain percentage]. To avoid sudden power jumps; S6. Based on the corrected air cutting parameters, the laser cutting head is driven to move along the loaded cutting trajectory, and the plate is cut using purified compressed air; The specific steps of step S6 are as follows: S61. Obtain the corrected air cutting parameters, wherein the corrected air cutting parameters include at least the corrected laser power, the corrected cutting feed speed, the corrected focal height, and the corrected compressed air pressure; S62. Send the corrected laser power to the power control interface of the fiber laser to control the laser to output a laser beam with the corresponding power; S63. Send the corrected cutting feed speed to the speed control interface of the servo motion system to control the laser cutting head to move along the cutting trajectory at the corrected speed; S64. Send the corrected focus height to the control interface of the cutting head height adjustment mechanism to adjust the distance between the cutting head and the surface of the material; S65. Send the corrected compressed air pressure to the control interface of the pressure regulating valve to adjust the air supply pressure; S66. Control the laser cutting head to move along the cutting trajectory under the combined action of the corrected laser power, cutting feed speed, focal height and compressed air pressure, and use purified compressed air to perform laser air cutting of the sheet metal; S7. Collect the cut image after the cutting is completed and judge the quality. If a defect is judged, record the process data and defect data of the current cutting process to the air cutting process database to trigger the iterative update of the process adaptive model. The specific steps of step S7 are as follows: S71. After laser cutting is completed, a visual image of the cut surface is acquired using an industrial camera. The industrial camera uses a CCD or CMOS sensor with a resolution of no less than 1280×1024 pixels, and is equipped with a ring light source or coaxial light source to provide uniform illumination and ensure image acquisition quality. S72. Preprocess the acquired cut image, wherein the preprocessing includes at least one of grayscale conversion, filtering and noise reduction, and edge detection; In step S72, grayscale conversion transforms the three-channel RGB image into a single-channel grayscale image using a weighted grayscale conversion formula:

[0023] in, Grayscale value , , These are the pixel values ​​for the red, green, and blue channels, respectively. The filtering and noise reduction adopts a multi-level filtering combination. First, a 3×3 median filter is used to remove discrete speckle noise, and then a 5×5 Gaussian filter operator with a standard deviation of 1.5 is used for smoothing. Edge detection uses the Canny edge detection operator, with a low threshold of 40 and a high threshold of 120. It combines non-maximum suppression and edge hysteresis connection strategies to extract the cut contour, slag outer contour and heat-affected zone boundary. S73. Extract defect feature parameters of the cut from the preprocessed image, wherein the defect feature parameters include at least one of slag height, cut penetration rate and heat-affected zone width; The slag hanging height is obtained in the following way: In the preprocessed cut image, the pixel coordinates of the theoretical reference bottom edge of the cut and the pixel coordinates of the highest position of the slag are extracted column by column along the cut normal direction. The pixel height of the slag at a single position is calculated. The pixel heights of all effective sampling positions are counted and the average value is obtained to get the average pixel height of the slag. The visual system was calibrated globally using a standard checkerboard calibration board. Under the same object distance, lens aperture and light source conditions as the actual detection, the visual calibration coefficient S was calculated by using the known physical size of the squares and the corresponding pixel width of the calibration board. Multiply the average pixel height of the dross by the visual calibration coefficient to convert it into the physical height of the dross. The cut penetration rate is obtained in the following way: The optimal grayscale segmentation threshold is adaptively solved using the Otsu method to segment the cut image into penetrated and non-penetrated regions. Laser-cut cuts are translucent, exhibiting low grayscale and dark characteristics, and are considered effective penetrated regions. Non-penetrated regions, areas with material adhesion, and residual slag are opaque, with grayscale characteristics similar to the substrate material and significantly higher grayscale values. Regions with pixel grayscale values ​​less than the optimal grayscale segmentation threshold are considered effective penetrated regions, while regions with pixel grayscale values ​​greater than or equal to the optimal grayscale segmentation threshold are considered non-penetrated regions. The pixel area of ​​the effective penetration region and the total pixel area of ​​the cut are calculated separately in the cut image, and the cut penetration rate is calculated using the following formula:

[0024] in, For incision penetration rate, To effectively penetrate the pixel area, This represents the total pixel area of ​​the cut. The width of the heat-affected zone is obtained in the following way: The cut-out melting contour extracted by the Canny edge detection operator is used as the inner boundary of the heat-affected zone, and the gray-level gradient abrupt change position at the junction of the substrate and the thermal oxidation discoloration area is used as the outer boundary of the heat-affected zone. Multiple sets of uniformly distributed sampling measurement lines are set in the incision image along the direction perpendicular to the edge of the incision. The pixel width between the inner and outer boundaries of the heat-affected zone is extracted line by line, and the average pixel width of all sampling positions is calculated. Multiply the average pixel width by the visual calibration factor to convert it into the actual width of the heat-affected zone; S74. Compare the defect feature parameters with the corresponding defect thresholds. If any defect feature parameter exceeds the corresponding defect threshold, it is determined that a defect exists. The defect thresholds are set differently based on the material and thickness of the plate: the threshold for slag height and the threshold for heat-affected zone width increase with the increase of plate thickness; the defect thresholds are different for different plate materials. S75. If a defect is determined to exist, the basic information of the plate material, the initial air cutting parameters, the corrected air cutting parameters, the multi-channel sensor signals and the defect data of the current cutting process are associated and recorded in the air cutting process database to trigger the iterative update of the process adaptive model. The iterative update of the triggering process adaptive model specifically includes: The basic information of the sheet metal, initial air cutting parameters, corrected air cutting parameters, multi-channel sensor signals and defect data of the current cutting process are associated and stored in the air cutting process database. When a new record in the database meets the preset triggering conditions, the accumulated data in the air cutting process database is used to perform incremental training or full retraining of the process adaptive model and update the model parameters. The preset triggering condition is at least one of the following conditions: The number of samples with the same material and thickness in the newly added records has reached the preset threshold. The newly added records contain samples with new materials or new thickness ranges; The current model's detection accuracy on the validation set has decreased by more than a preset percentage compared to the previous iteration; When training the process adaptive model using accumulated data from the air cutting process database, a sample weighting strategy is adopted: recent samples are given higher weights, and the weights of historical samples decrease as the storage time increases; at the same time, samples of working conditions with a sample count below a preset threshold are given higher weights than samples of normal working conditions. The air cutting process database is a relational database, containing: The cutting process parameter table is used to store the basic information of the sheet material, the initial air cutting parameters, and the corrected air cutting parameters for each cutting process; A cutting defect sample data table is used to store the defect characteristic parameters and defect judgment results of each cutting process. The cutting defect sample data table is associated with the cutting process parameter table through the sample number. The model training parameter record table is used to store the number of training samples, model hyperparameters, training loss values, and inference accuracy for each iteration of the process adaptive model training.

[0025] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0026] like Figure 2 As shown, the following are embodiments of the intelligent air-cutting plate processing system provided in this disclosure. This system and the intelligent air-cutting plate processing methods of the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the intelligent air-cutting plate processing system, please refer to the embodiments of the intelligent air-cutting plate processing methods described above.

[0027] The system includes: The board material information acquisition module is used to acquire basic information about the board material to be processed, which includes at least the board material and the board thickness. The process parameter matching module is used to call the air cutting process database and match the initial air cutting parameters according to the basic information of the plate. The initial air cutting parameters include at least laser power, cutting feed speed, focal height and compressed air pressure. The compressed air supply and purification module is used to start the air compressor supply device, purify the compressed air, and use the purified compressed air as the auxiliary gas for laser cutting. The multi-signal sensing module is used to load the cutting trajectory and execute the cutting, and to collect multiple sensing signals in real time during the cutting process. The multiple sensing signals include plasma optical signals, the distance signal between the cutting head nozzle and the plate, and gas pressure feedback signals. The intelligent parameter control module is used to input multi-sensor signals into the trained process adaptive model. The process adaptive model outputs at least one of laser power compensation, feed speed compensation and air pressure compensation, and uses the compensation to dynamically correct the current air cutting parameters. The laser cutting execution module is used to drive the laser cutting head to move along the loaded cutting trajectory based on the corrected air cutting parameters, and to cut the sheet material using purified compressed air; The vision inspection module is used to acquire images of the cut surface after the cutting is completed and to determine the quality. If a defect is determined, the process data and defect data of the current cutting process are recorded in the air cutting process database to trigger the iterative update of the process adaptive model.

[0028] This embodiment achieves intelligent cutting throughout the entire process through the interactive collaboration of the sheet material information acquisition module, process parameter matching module, compressed air supply and purification module, multi-signal sensing module, intelligent parameter control module, laser cutting execution module, and visual inspection module. This overcomes the sensitivity to fluctuations in operating conditions and enables low-cost, high-quality, and highly stable cutting.

[0029] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for processing sheet metal using intelligent air cutting, characterized in that, Includes the following steps: S1. Obtain the basic information of the board to be processed, wherein the basic information of the board includes at least the board material and the board thickness; S2. Call the air cutting process database and match the initial air cutting parameters according to the basic information of the plate. The initial air cutting parameters include at least the laser power, cutting feed speed, focal height and compressed air pressure. S3. Start the air compressor supply device to purify the compressed air and use the purified compressed air as the laser cutting auxiliary gas; S4. Load the cutting trajectory and execute the cutting, and collect multiple sensor signals in real time during the cutting process. The multiple sensor signals include plasma light signals, the distance signal between the cutting head nozzle and the plate, and gas pressure feedback signals. S5. Input the multi-sensor signals into the trained process adaptive model, the process adaptive model outputs at least one of laser power compensation, feed speed compensation and air pressure compensation, and uses the compensation to dynamically correct the current air cutting parameters. S6. Based on the corrected air cutting parameters, the laser cutting head is driven to move along the loaded cutting trajectory, and the plate is cut using purified compressed air; S7. Collect the cut image after the cutting is completed and judge the quality. If a defect is judged, record the process data and defect data of the current cutting process to the air cutting process database to trigger the iterative update of the process adaptive model.

2. The processing method for intelligent air-cutting sheet metal according to claim 1, characterized in that, The specific steps of step S1 are as follows: S11. Input the material and thickness of the board to be processed through the human-computer interaction interface or import it through the upstream system; S12. Obtain surface condition information of the board material through a visual sensor or laser scanner, wherein the surface condition information of the board material includes at least one of surface roughness, oxide layer thickness and surface flatness; S13. Combine the material, thickness, and surface condition of the board into a basic information vector of the board.

3. The processing method for intelligent air-cutting sheet metal according to claim 2, characterized in that, The specific steps of step S2 are as follows: S21. Using the basic information of the sheet metal as the search criteria, query the air cutting process database; the air cutting process database stores several historical process records, each historical process record includes the sheet metal material, sheet metal thickness, sheet metal surface condition and corresponding optimal air cutting parameters, the optimal air cutting parameters include at least laser power, cutting feed speed, focal height, compressed air pressure and nozzle type; S22. Retrieve historical process records in the air cutting process database that match the current sheet material and sheet thickness; If multiple historical process records are matched, the air cutting parameters corresponding to the historical process record that is closest to the current surface condition of the board are selected as the initial air cutting parameters. If no matching record is found, the default parameter group will be used as the initial air cutting parameters. S23. Output the initial air cutting parameters obtained from the matching.

4. The processing method for intelligent air-cutting sheet metal according to claim 1, characterized in that, The specific steps of step S3 are as follows: S31. Start the air compressor to generate compressed air and send the compressed air into the air tank for buffering and pressure stabilization; S32. After buffering and stabilizing, the compressed air is sequentially passed into an oil-water separator for preliminary oil-water separation, into a refrigerated dryer for drying, and into a multi-stage filter with progressively increasing filtration precision for filtration and purification; the multi-stage filter includes at least a primary filter and a secondary filter, wherein the filtration precision of the secondary filter is higher than that of the primary filter. S33. The pressure of the purified compressed air is stabilized by a pressure regulating valve to control the dew point and oil concentration of the purified compressed air within preset values; S34. The purified compressed air after pressure stabilization is delivered to the laser cutting head as auxiliary gas through the air supply pipeline, and the real-time pressure value of the air supply pipeline is output as a gas pressure feedback signal.

5. The processing method for intelligent air-cutting sheet metal according to claim 4, characterized in that, The specific steps of step S4 are as follows: S41. Read the G-code file of the cutting trajectory of the part to be processed and parse it to obtain the preset motion trajectory of the laser cutting head; S42. Start laser cutting processing and simultaneously activate multiple sensors for sensing; S43. The intensity of the plasma light signal generated during the cutting process is acquired in real time by a plasma sensor, wherein the intensity of the plasma light signal characterizes the melting and vaporization state of the material in the cutting area; S44. The capacitance value between the cutting head and the plate is detected in real time by a capacitance height sensor, and the real-time distance signal of the cutting head is calculated based on the capacitance value. The real-time distance signal represents the following height between the nozzle of the cutting head and the surface of the plate. S45. Real-time acquisition of gas pressure feedback signals in the gas supply pipeline via a gas pressure transmitter; S46. The real-time acquired plasma optical signal, the distance signal between the cutting head and the plate, and the gas pressure feedback signal are combined into a multi-channel sensing signal vector and transmitted to the intelligent parameter control module at preset sampling intervals as input to the process adaptive model described in step S5.

6. The processing method for intelligent air-cutting sheet metal according to claim 5, characterized in that, The specific steps of step S5 are as follows: S51. Construct a process adaptive model, wherein the process adaptive model is a neural network model trained with historical cutting process data, wherein its input layer receives the multi-channel sensor signals, and its output layer outputs laser power compensation, feed speed compensation and air pressure compensation. S52. Input the multi-channel sensor signals that are collected and transmitted in real time into the process adaptive model. The process adaptive model performs forward inference calculation and outputs at least one of the laser power compensation, feed speed compensation and air pressure compensation at the current moment. S53. The compensation amount output by the process adaptive model is superimposed on the current air cutting parameters to dynamically correct the current air cutting parameters: in, This is the corrected real-time laser power. The current laser power, This is the laser power compensation amount; This is the corrected real-time cutting feed rate. This represents the current cutting feed rate. This is the feed rate compensation amount; This is the corrected real-time compressed air pressure. The current compressed air pressure, This is the air pressure compensation amount; S54. Output the corrected air cutting parameters.

7. The processing method for intelligent air-cutting sheet metal according to claim 6, characterized in that, In step S5, geometric features on the cutting trajectory are also identified, including at least one of corners, small holes, and narrow slits; When the geometric features are identified, the corrected air cutting parameters are adjusted for energy compensation. The energy compensation regulation specifically includes: Based on the type and size of the geometric features, the corresponding power compensation coefficient and speed reduction coefficient are determined respectively; The current laser power is corrected using a power compensation factor: in, This represents the laser power after energy compensation. The current laser power, The power compensation coefficient corresponding to the geometric feature type; The current cutting feed rate is corrected using a speed reduction coefficient: in, This refers to the cutting feed rate after energy compensation. This represents the current cutting feed rate. This is the velocity deceleration coefficient corresponding to the geometric feature type.

8. The processing method for intelligent air-cutting sheet metal according to claim 6, characterized in that, The specific steps of step S6 are as follows: S61. Obtain the corrected air cutting parameters, wherein the corrected air cutting parameters include at least the corrected laser power, the corrected cutting feed speed, the corrected focal height, and the corrected compressed air pressure; S62. Send the corrected laser power to the power control interface of the fiber laser to control the laser to output a laser beam with the corresponding power; S63. Send the corrected cutting feed speed to the speed control interface of the servo motion system to control the laser cutting head to move along the cutting trajectory at the corrected speed; S64. Send the corrected focus height to the control interface of the cutting head height adjustment mechanism to adjust the distance between the cutting head and the surface of the material; S65. Send the corrected compressed air pressure to the control interface of the pressure regulating valve to adjust the air supply pressure; S66. Control the laser cutting head to move along the cutting trajectory under the combined action of the corrected laser power, cutting feed speed, focal height and compressed air pressure, and use purified compressed air to perform laser air cutting of the sheet metal.

9. The processing method for intelligent air-cutting sheet metal according to claim 8, characterized in that, The specific steps of step S7 are as follows: S71. After laser cutting is completed, a visual image of the cut surface is captured using an industrial camera; S72. Preprocess the acquired cut image, wherein the preprocessing includes at least one of grayscale conversion, filtering and noise reduction, and edge detection; S73. Extract defect feature parameters of the cut from the preprocessed image, wherein the defect feature parameters include at least one of slag height, cut penetration rate and heat-affected zone width; S74. Compare the defect feature parameters with the corresponding defect thresholds. If any defect feature parameter exceeds the corresponding defect threshold, it is determined that a defect exists. S75. If a defect is determined to exist, the basic information of the plate material in the current cutting process, the initial air cutting parameters, the corrected air cutting parameters, the multi-channel sensor signals and the defect data are associated and recorded in the air cutting process database to trigger the iterative update of the process adaptive model.

10. A processing system for intelligent air-cutting sheet metal, characterized in that, include: The board material information acquisition module is used to acquire basic information about the board material to be processed, which includes at least the board material and the board thickness. The process parameter matching module is used to call the air cutting process database and match the initial air cutting parameters according to the basic information of the plate. The initial air cutting parameters include at least laser power, cutting feed speed, focal height and compressed air pressure. The compressed air supply and purification module is used to start the air compressor supply device, purify the compressed air, and use the purified compressed air as the auxiliary gas for laser cutting. The multi-signal sensing module is used to load the cutting trajectory and execute the cutting, and to collect multiple sensing signals in real time during the cutting process. The multiple sensing signals include plasma optical signals, the distance signal between the cutting head nozzle and the plate, and gas pressure feedback signals. The intelligent parameter control module is used to input multi-sensor signals into the trained process adaptive model. The process adaptive model outputs at least one of laser power compensation, feed speed compensation and air pressure compensation, and uses the compensation to dynamically correct the current air cutting parameters. The laser cutting execution module is used to drive the laser cutting head to move along the loaded cutting trajectory based on the corrected air cutting parameters, and to cut the sheet material using purified compressed air; The vision inspection module is used to acquire images of the cut surface after the cutting is completed and to determine the quality. If a defect is determined, the process data and defect data of the current cutting process are recorded in the air cutting process database to trigger the iterative update of the process adaptive model.

Images